You will be able to: › Explain what is meant by an expert system and describe its components and applications.
Society is becoming more complex. More and more information is becoming available. Systems need to be designed to deal with information so that it can be found more easily by humans.
Alternative name is knowledge-based system. Consider: › The problem of diagnosing a patient’s illness. › Doctors can break down the symptoms and use their knowledge and possibly tests on the patient to try and figure out what the probably illness is. › Sometimes it takes a team of doctors to work out what the problem is.
Now consider: › A computer that has all the information stored about every illness known to man. › It stores information such as symptoms and treatments. › As a patient describes their symptoms the doctor can input these into the system and the computer displays all the illnesses that match up with the symptoms. › Using the new facts the doctor can eliminate tests that do not need to be carried out and it lowers the risk of mis- diagnosing patients.
Expert systems usually contain the knowledge of a large number of specialists. They can search through information very quickly much faster than a human. Less likely to forget details about rare diseases, unlike a human. Could be used by people to decide if they need to go to the doctors or not. › A bit like the family doctor books!
Can sometimes be expensive to be set up. They still rely on accurate information being put into the system. One mistake could be costly! Useless if asked questions outside their knowledge bank. Can people really trust a machine??
Will compose of a natural language interface. The ultimate aim is to be able to hold a conversation with a computer. The Turning Test › Alan Turning developed a simple test to see if a computer possessed intelligence. › People had to work out if they were talking to another person or a computer. › Most people realised that it was a computer.
Symptoms tend to exist for many types of illnesses. It is unlikely that the expert system will find one illness that meets all the criteria. Multiple responses may be given for a headache, high temperature and cough for example: › Common Cold = 61.78% chance › Influenza = 38.21% chance › Bubonic plague = 0.01% chance
There are four parts to an expert system: › The knowledge base › The rule base › The inference engine › The user interface
The knowledge base The rule base The inference engine The inference engine The user interface The user interface user
This sections holds the facts. It will hold information related to the topic that the expert system is intended for.
This section holds rules that need to be applied to the knowledge base. For example: › IF a toe nail is swollen › AND the nail is black OR there is puss › THEN it could be an in-growing toenail.
This is the brains of the system. It is where all the clever bits happen. It separates the user interface from the clever bit of the system. It allows different user interfaces to be created to suit different needs…the knowledge base remains the same…its only the interface that changes.
The user will need an interface through which they can interact with the system. The interface may display closed questions which will result in different responses being displayed. One day open ended questions might be accepted.
For each type you will need to be able to describe an example. › Diagnostic systems › Advice-giving systems › Decision making systems
Asks a question to narrow down the number of results from the knowledge bank. › Classic example is the medical system we have already covered. Consider the following diagram and note how the knowledge bank gets smaller. Knowledge bank Yes No Yes No Yes No Yes No Yes No Yes No Yes No Question 1Question 2Question 3
A stock monitoring system is a good example of this. It monitors the level of stock and then when levels drop too low it will advise that stock may need to be reordered. This would then be done manually if they want to continue stocking that particular product.
A more advanced system compared to advice-giving systems. Taking the stock example again: › Stock can be reordered automatically when conditions meet specified requirements such as: Stock level falls below a set level Time of year and market trends Number of sales in recent days, weeks, months etc Etc…
1. Define a ‘knowledge-based system’ 2. Describe the parts that make up a knowledge-based system, using examples where you can. 3. What are the characteristics of an expert system? 4. State three types of knowledge-based system and give examples of each. 5. What is the essential difference between an advice- giving system and a decision-making system? 6. Describe the pros and cons of introducing a medical expert system into a third world country in place of a doctor.